[PSUBS-MAILIST] Optical Oxygen Sensors

Tue Aug 20 00:42:39 EDT 2019

Alan, not to discourage, but think carefully about the potential pitfalls of implementing sensor voting logic. This is unquestionably an improvement over a single sensor, but should not be considered a panacea. Such systems often do not account for common failure modes which can cause two, or all sensors to read erroneously, and possibly in concert providing no indication of a problem. (One such example is condensation on the sensor membranes, which can affect sensitivity similarly on all similar sensors). As an alternative to voting, you might want to consider active periodic sensor verification. A bit more complicated mechanically, but such an embodiment would cycle through your installed sensors, isolating each one in turn from the averaging algorithm to perform a verification for continuous sensor health monitoring. Consider that you have N sensors each installed directly in the oxygen injection path, which is valved appropriately to direct the oxygen flow across the face of one selected sensor, while the remaining N-1 sensors are exposed only to the cabin air. The idea is that the directed flow of dry oxygen across the selected sensor membrane will discourage condensation on the membrane, while simultaneously inducing a spike in the oxygen reading on that sensor when the injection occurs. The indicated oxygen level on the spiked sensor in the flow path is used as a measure of sensor health (because this can be tracked), while the remaining sensors which are not in the injection path are averaged to produce your representative cabin oxygen reading. Once the indicated oxygen level on the spiked sensor returns to normal (I.e. to within some threshold of the active average), it is returned to the averaging group, and the next sensor in line is similarly isolated before the next oxygen injection. This is technically superior to simply implementing a voting algorithm, because it isn't making any assumptions about sensor performance. A more advanced implementation of this which would be particularly useful in a slow system where the oxygen injection events are quite infrequent, would be to weight the average cabin oxygen readings according to how recently each sensor in the averaging group had been verified.

Sean

Sent from ProtonMail mobile

-------- Original Message --------
On Aug. 19, 2019, 21:16, Alan via Personal_Submersibles wrote:

> Thanks Cliff,
> I just noticed the second half of your email below the product picture :(
> I like that O2 sensor you linked to & have noted the 10-15mv specs so I
> can write code & do the electronics for it, but buy them when I have finished
> the rest of the sub. If I bought them first they may expire before I finish.
> I am wanting to do similar to you except use 3 X O2 sensors. I am going to
> compare the readings & go with the average of the nearest 2, as in a rebreather
> system.
> One reason is it is a 1 person sub & I want to be able to send inexperienced
> people down in it who may panic if alarms went off for a sudden sensor
> failure & also may not recognise what was going on. I also want to signal which
> sensor is going out of range & monitor it.
> Alan
>
> On 20/08/2019, at 1:59 AM, Cliff Redus via Personal_Submersibles <personal_submersibles at psubs.org> wrote:
>
>> Alan, I am with Jon on the span for the O2 sensor.  Granted under NOP a 0-25% span for O2 percent is fine but 0-100% span is handy to detect if you have HP O2 bleed into the cabin.  On one dive weekend a few years ago,   when I got ready to dive the next day I powered up the boat and a high O2 alarm sounded in the boat.  After checking I found that I had left the HP O2 blocking valve on my O2 tank open over night with the hatch secured.  With my system it is possible to have a very low flow bleed into the cabin. The cabin atmosphere had risen to 27% O2.  I got out my handy leaf blower and aired out the cabin and proceeded to dive.
>>
>> I use the [Max-250 | Maxtec](https://www.maxtec.com/product/sensing/fio2/max-250/)
>>
>> https://www.maxtec.com/product/sensing/fio2/max-250/
>>
>> Max-250 | Maxtec
>>
>> Replacement oxygen sensor for Maxtec OM-25, Fluke VT Mobile,
>>
>> with a custom PCB that converts this 10-15mV signal into a 0-5V signal I use for analog input on my PLC.  I find you have to replace this element every 3-4 years.  It comes factory calibrated.  For field calibration, I added some ladder logic to PLC that lets me add a bias error correction to sensor by comparing what the sensor is measuring to the air it is measuring in the cabin when the hatch is open.  I find this works quite well.  Normal air has 20.95% O2, if I look at my life support display on my boat and find it much off from this, I hit the calibration button and difference between what it is reading and 20.95% is stored as a bias error correction calibration constant.
>>
>> The accuracy of Max-250 sensor is +/- 1% full scale so the benefit of going to a O2 sensor with a 0-25% span even with with a +/- 2% would be a more accurate measurement because of the span.  Having said that, I have had no issues with 0-100%, +/-1% full scale sensor.
>>
>> The nice part about having the output from the O2 sensor go through the PLC is that I can have the PLC sound an alarm and lock out diving the boat if the O2 measurement is high or low.
>>
>> For my boat I sound a high O2 alert if O2 is > 23% and a low O2 alert if the O2 is <18%.  These alarm levels have worked pretty well.
>>
>> Cliff
>>
>> On Monday, August 19, 2019, 08:15:04 AM CDT, Jon Wallace via Personal_Submersibles <personal_submersibles at psubs.org> wrote:
>>
>> Hi Alan,
>>
>> I still think the major advantages of the optical sensor is the digital output and it's expected life of 5 years.  But I have a hard time getting past the 25% scale because there's a lot of "head room" between 25-100% and you just don't know where you are in that range.  My concern is not from a biological perspective since short term high levels of O2 are not an issue at 1-ATM, but from an environment one...high levels of 02 start representing a real fire danger.
>>
>> Calibration in free air is easiest and why handheld (mobile) sensors are convenient.  I'm not sure I would trust just opening the hatch to get fresh air in the vessel, unless you forced fresh air inside with a fan or something.  Getting into an open air environment would be best.
>>
>> Jon
>>
>> On Monday, August 19, 2019, 05:58:02 AM EDT, Alan via Personal_Submersibles <personal_submersibles at psubs.org> wrote:
>>
>> Jon,
>> am having second thoughts about the optical O2 sensor.
>> I was given data that said calibration was needed after a while because of a drift
>> caused by temperature fluctuations.
>> Although calibration would only need to be done every year or two, I would
>> need to write a program for that. And I would need to write a program for
>> calibration of galvanic O2 sensors if I went with them. So although the optical
>> sensors are factory calibrated, this is not going to be that much of an advantage
>> to me. They would also be more difficult to source than the standard sensor.
>> The optical sensors only have the 0-25% O2 range but I can't imagine going over
>> 25%. The optical sensor has better accuracy, as all the information I have seen
>> on the galvanic sensors say they have an accuracy of +/- 2% on full scale.
>> I wonder what procedures Psubbers take with regards to checking calibration
>> of their O2 sensors. I am thinking a best practice would be to leave the hatch
>> open before a dive & check that the O2 reading corresponds to the standard
>> level of O2 in air.
>> Cheers Alan
>>
>> _______________________________________________
>> Personal_Submersibles mailing list
>> Personal_Submersibles at psubs.org
>> http://www.psubs.org/mailman/listinfo.cgi/personal_submersibles
>
>> _______________________________________________
>> Personal_Submersibles mailing list
>> Personal_Submersibles at psubs.org
>> http://www.psubs.org/mailman/listinfo.cgi/personal_submersibles
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.whoweb.com/pipermail/personal_submersibles/attachments/20190820/a63e685f/attachment-0001.html>